Impact of network randomness on multiple opinion dynamics

TL;DR

This paper investigates how the randomness in network connections influences the final distribution of opinions in models of opinion dynamics, revealing that increased randomness promotes consensus and can sustain indecision.

Contribution

It compares two opinion dynamics rules on Watts-Strogatz networks, highlighting how network randomness affects consensus, coexistence, and indecision in opinion formation.

Findings

Increased network randomness favors consensus.

Plurality rule dynamics can sustain undecided nodes.

Maximum indecision occurs in the small-world network region.

Abstract

People often face the challenge of choosing among different options with similar attractiveness. To study the distribution of preferences that emerge in such situations, a useful approach is to simulate opinion dynamics on top of complex networks, composed by nodes (individuals) and their connections (edges), where the state of each node can be one amongst several opinions including the undecided state. We use two different dynamics rules: the one proposed by Travieso-Fontoura (TF) and the plurality rule (PR), which are paradigmatic of outflow and inflow dynamics, respectively. We are specially interested in the impact of the network randomness on the final distribution of opinions. For that purpose, we consider Watts-Strogatz networks, which possess the small-world property, and where randomness is controlled by a probability $p$ of adding random shortcuts to an initially regular network. Depending on the value of $p$, the average connectivity $\langle k \rangle$, and the initial conditions, the final distribution can be basically (i) consensus, (ii) coexistence of different options, or (iii) predominance of indecision. We find that, in both dynamics, the predominance of a winning opinion is favored by increasing the number of reconnections (shortcuts), promoting consensus. In contrast to the TF case, in the PR dynamics, a fraction of undecided nodes can persist in the final state. In such cases, a maximum number of undecided nodes occurs within the small-world region, due to ties in the decision group.